Abstract
This paper develops an economic and social capital model linking single mothering in poverty to adolescent violence. Our model focuses on bonding social capital within parent-child relationships, negative social capital in delinquent peer groups, and bridging social capital residing in youths’ friendship networks. Our research is the first to consider that the family experiences of adolescents’ peers affect adolescent violence. We test hypotheses using the Add Health, finding that peer networks are a source of bridging social capital through which collective parenting helps explain youth violence as well as the links between family structural (dis)advantages and youth violence.
Keywords
Introduction
The proportion of two parent households has declined since the early 1970s, whereas the proportion of single mother households has increased, leveling off at fairly high levels in recent years. Today, nearly a quarter of children under the age of 18—a total of 16.4 million—are being raised by a single mother. Sixty-nine percent of their peers—a total of 51 million—are being raised in two-parent households (U.S. Census Bureau, 2019). While two-parent families remain the norm, there are substantial numbers of young people coming of age in homes headed by single mothers. Among children living in single-mother families, 39% experience poverty as compared to only 8% of those residing in two-parent families (U.S. Census Bureau, 2019). These demographic patterns have fueled debate about whether and how family structure influences childhood outcomes, including adolescent offending (Edin, 2004; McLanahan & Sandefur, 1994; Wilson, 2012).
The link between single-mother households and adolescent offending has amassed considerable research attention (Amato, 2001; Wong, 2017). Nevertheless, more research is needed to unpack how poverty and access to social resources within and outside the family influence delinquency across family structures. Prior research has focused on understanding the contexts within which single-mother families are most consequential for adolescent offending or on articulating the processes through which single-mothering is linked to delinquency (e.g., De Coster, 2012; Kierkus & Hewitt, 2009). We draw on both bodies of research to develop an economic and social capital model specifying the structural contexts and cultural processes that render single mothering relevant for violent delinquency. Our model specifies family poverty as a contextual factor that informs the relationship between single-mother families and adolescent violence. Our articulation of the processes through which poverty and single mothering combine to predict violence focuses on three forms of social capital—bonding, negative, and bridging (Putnam, 2000).
Social capital consists of relationships between individuals that allow for the communication of information and norms, as well as sanctions for behaving outside group norms (Coleman, 1988). Bonding social capital refers to intergenerational intrafamilial relationships with an emphasis on socialization within the parent-child relationship. Research emphasizes the relevance of bonding social capital as it plays out through parental supervision, monitoring, and the fostering of emotional bonds with children (Dufur et al., 2015). Negative social capital accrues in networks that support and facilitate offending, including delinquent peer groups (McCarthy & Hagan, 1995). And, bridging capital focuses on intergenerational interfamilial relationships (Hoffmann & Dufur, 2018).
Our paper particularly focuses on bridging social capital embedded in peer networks. In doing so, we introduce the concept of network parenting, or the shared parenting potential in adolescent friendship networks as measured by characteristics of the families of youths’ friends. Specifically, we examine aspects of friendship networks that reflect the family structure of network members (the proportion of adolescents in the friendship network living in poor single-mother households and the proportion living in nonpoor households that are not headed by single mothers) and aspects of networks that reflect family process (the proportion of adolescents in the network who have close relationships with their parents). Wedding sociological and criminological theories, we specify hypotheses about how and why poverty and network parenting in friendship groups matter for violent delinquency as well as for the relationship between impoverished single mothering and violence. We test our hypotheses using the National Longitudinal Study of Adolescent to Adult Health (Add Health).
One of the contributions of our research is our focus on poverty as core to understanding links between single-mothering and violent delinquency. In doing so, we demonstrate that it is not single-mothering per se that is related to social capital resources and adolescent violence; rather it is single-mothering within the context of poverty that matters. A second contribution is our focus on network parenting as relevant for understanding the relationship between living in a poor single-mother home and violent delinquency. Prior research has focused on how parents control teens in their own household (Laub & Costello, 2020) or on the ability of community networks to control local teenagers (Kubrin & Mioduszewski, 2019). These foci parallel the bonding and bridging capital in social capital theory. We extend prior work by considering friendship networks as a key source of bridging social capital. Our research is the first to consider that the likelihood of violent delinquency is associated with the proportion of youths’ friends who live in poor single-mother homes and with the proportion of friends who are emotionally bonded to their parents. Our findings highlight the importance of the friendship group as a site for bridging social capital or collective parenting. Specifically, we highlight the key role of the family experiences of youths’ friends, above and beyond the family experiences of youths themselves.
Poverty, Single-Mother Families, Social Capital, and Adolescent Offending
Many studies have focused on the relationship between single-mother families and adolescent offending (for reviews, see Amato, 2001; Wells & Rankin, 1991). Explanations for the association focus on the economic strains accompanying single mothering and on differences in family processes across single-mother and two-parent households (De Coster, 2012; Leiber et al., 2009). Single-mother families are nearly five times more likely to experience poverty than two-parent families and twice as likely as single-father families to live below the poverty line (U.S. Census Bureau, 2019). Structural perspectives maintain that poverty explains elevated rates of delinquency in single-mother families because of the stresses associated with raising children while struggling to make ends meet (Amato, 2001; Leiber et al., 2009). Qualitative studies show that poverty can be psychologically and physically punishing for single mothers as they toil to piece together insufficient and precarious resources to enable their families to get by (Edin & Kefalas, 2011; Edin & Shaefer, 2015). These findings from qualitative studies on how poverty affects families provide texture to findings from quantitative studies rooted in family process models.
Family process models propose that parenting practices and parent-child relationships are central in linking the stresses of single-mothering and poverty to delinquency and violence (Deutsch et al., 2012; McLanahan, 2009). The argument is that the significant stresses accompanying single mothering and poverty add to the burden of parenting, making it difficult for poor single mothers to devote as much time as non-poor mothers or two-parent families to supervising and monitoring children or to maintaining strong emotional bonds with children (Daryanani et al., 2016; Kendig & Bianchi, 2008). Poorer supervision, lower parental monitoring, and weakened bonds between parents and children, in turn, are associated with higher risk of delinquency and violence (Hoffmann & Dufur, 2018). This is a central tenet of social control and social capital theories of adolescent offending (Hirschi, 1969; Hoffmann & Dufur, 2018). Ultimately, the family process model posits that bonding social capital in the forms of supervision, monitoring, and emotional bonds is undermined in poor single mother families, which places adolescents at risk of engaging in delinquency and violence.
Studies focusing on economic strains and family processes often report that the association between single-mother households and offending is not explained fully by poverty and parenting (Demuth & Brown, 2004; Rebellon, 2002). In other words, intrafamilial dynamics are insufficient for understanding why residing in a single-mother home increases adolescent offending. This is unsurprising, given that adolescents are embedded in multiple social contexts, including peer and community settings that are relevant for understanding their development and behaviors (Bronfenbrenner & Evans, 2002).
Two streams of research address how additional social contexts help explain the link between single-mother families and delinquency. First, some scholars emphasize the role of delinquent peers in promoting offending (Akers & Jennings, 2019). These researchers show that the stresses associated with poverty and single mothering negatively affect parental supervision, monitoring, and bonds with children, which in turn have consequences for peer associations and delinquency (De Coster, 2012; Deutsch et al., 2012). Weakened supervision and reduced monitoring affect peer associations because children’s friendships are unregulated, allowing youths to associate with delinquent peers (Matsueda,1982). Further, adolescents who have weak bonds with their parents are not motivated to consider their parents’ preferences when selecting friendship groups and often navigate toward delinquent peers (De Coster, 2012). Delinquent peers, in turn, encourage offending by teaching values favoring delinquency and offering status, identity, and friendship rewards for delinquent behaviors (Heimer & De Coster, 1999; Matsueda, 1982). In other words, delinquent peer groups are a source of negative social capital (Haynie & Payne, 2006).
Research that includes both family and peer processes accounts more fully for the relationship between single mothering and delinquency than research that focuses solely on family processes (De Coster, 2012; Deutsch et al., 2012). Such research has relied on indirect measures of peer delinquency wherein adolescents report how many of their friends break the law. This can result in overestimates of the relationship between delinquent friendships and delinquency because adolescent respondents may project their own behaviors onto their peers (Weerman & Smeenk, 2005). Our research goes beyond prior studies of family structure and delinquency by adopting a social network approach that maps out friendship networks and incorporates friends’ own reports of their offending.
A second stream of research focuses on the contexts within which single mothering is most consequential for adolescent offending. One line of work in this stream focuses on the socio-economic context in which single mothers parent their children, positing that the stresses of single mothering are amplified when mothers are struggling to provide for their family. Some research finds that adolescents residing in poor single-mother homes are at greater risk for misbehavior than those in single-mother homes with economic resources (Goldstein, 1984; Touliatos & Lindholm, 1980). Other research, however, fails to find this pattern (Kierkus & Hewitt, 2009). One explanation for mixed support is that researchers use different measures of family socioeconomic status and focus on different types of delinquency. Studies that include dichotomous measures of family poverty rather than continuous measures of income and those that focus on serious rather than petty delinquency are more likely to uncover links between family socioeconomic resources and delinquency (Elliott & Ageton, 1980; Farnworth et al., 1994).
An additional explanation for null findings is that the intersection of single mothering and economic disadvantage may operate primarily at the community level (Hoffmann, 2006; Kierkus & Hewitt, 2009). That is, the social and economic disadvantages of the communities in which families reside may be more important for conditioning the effect of family structure on adolescent offending than the economic disadvantage of families themselves. Single-mother families often live in communities characterized by high concentrations of poverty and single-mother households (De Coster et al., 2006). Research shows that rates of delinquency and violence are higher in communities with high proportions of single mothers, presumably because there are fewer adults available to offer support, monitoring, and supervision for their own children and for children in the community (Kubrin & Mioduszewski, 2019). In other words, disadvantaged communities experience a shortage of bridging social capital (Sampson, 2012). Although collective supervision has received primary empirical attention in the literature on communities and crime, Bursik (1988) emphasizes that community parenting entails not only intervening in problem behaviors through supervision but also includes the cultivation of intergenerational relationships that encourage adolescents to refrain from offending even when they are not being directly supervised by community residents (Swaroop & Morenoff, 2006).
Some scholars suggest that the collective supervision made possible by community bridging capital in advantaged areas can compensate for weak bonding capital within families (Hay et al., 2006; Weatherburn & Lind, 2001). On the flip side, positive parenting within families can be undermined in disadvantaged communities characterized by low levels of collective parenting (Anderson, 1999; De Coster et al., 2006). Thus, adolescents living with single mothers in disadvantaged communities with many other single-mother families may be at greater risk for delinquency than their counterparts living with single mothers in advantaged communities with many intact families. Though often discussed, this multiplicative hypothesis has received scant empirical attention. Hoffmann’s (2006) empirical assessment shows that community disadvantage does not condition the relationship between family structure and delinquency. As such, we focus on the more intimate and immediate context of peer networks.
Bridging Social Capital and Network Parenting
Extrapolating from community context arguments, we argue that the family structure and poverty of adolescents’ peer networks are relevant for violence. Despite recognizing delinquent peers as a source of offending, research on family structure and offending has not unpacked how the characteristics of peer networks shape the relationship between poor single-mother households and delinquency. We propose that adolescents who are embedded in networks of friends from poor single-mother families are less likely than other adolescents to be controlled and supported through relationships with their friends’ parents. By comparison, adolescents embedded in networks of friends from households that are neither poor nor headed by single mothers are more likely to experience social control through relationships with their friends’ parents. Thus, adolescents gain access through their peer groups to network parenting or bridging social capital. Our argument includes both consideration of structural aspects of networks (friends’ residence in impoverished single-mother families) and process elements (intergenerational relationships within the networks). The dual operation of family structural and process resources in friendship networks has not been considered previously in attempts to understand how single mothering and poverty combine to influence offending.
Examining the characteristics of friendship networks is particularly relevant to the relationship between single-mothering and delinquency when we consider that adolescents from impoverished, single-mother families often have friends from similarly situated families, given the clustering of poor single-mother families in disadvantaged communities. Moreover, adolescents from poor single-mother homes may seek one another out because they find comfort among peers with similar family backgrounds and struggles. In short, understanding how single motherhood and poverty coalesce to inform delinquency requires considering the attributes of friends’ families.
Synthesis and Hypotheses
We propose a model linking the confluence of single-mothering and poverty to violent delinquency through bonding, negative, and bridging social capital. In doing so, we introduce the idea that the family experiences of youths’ friends are key to understanding adolescent violence, above and beyond the family experiences of youths themselves. This proposition goes beyond research that has focused on the concentration of poor single-mother families in communities to focus on the more intimate social setting of friendship groups. By considering friendship groups as a potential source of bridging social capital, we assess how friendship networks of shared family structural and relational (dis)advantages affect violent delinquency, as well as help explain why living in a poor single-mother household increases youth violence.
We propose that it is not single mothering per se that is key to understanding access to social capital resources and adolescent violence. Instead, we argue that it is critical to consider the socioeconomic context in which single mothers parent children in order to understand linkages between single mothering, social capital, and juvenile violence.
Hypothesis 1: Adolescents from poor single-mother families are more likely to engage in violent delinquency than those from other families, including single mother families that are not poor.
Family process models link family disadvantage to delinquency through parenting practices, focusing on the bonding social capital inherent in intergenerational relationships within families. According to these models, the stresses associated with single mothering and poverty add to the burden of parenting, making it difficult for mothers to find the time, patience, and energy to supervise, monitor and maintain strong bonds with their adolescents. Parental supervision, monitoring, and emotional bonds, in turn, are associated with violent delinquency.
Hypothesis 2:
Research suggests that along with influencing violence directly, bonding social capital is linked to violence indirectly through peer relationships. Adolescents who are not well supervised, poorly monitored, and weakly bonded to their parents are at increased risk of associating with delinquent peers. This is because adolescents who are unsupervised, unmonitored, and weakly bonded to their parents are free to choose friendships with little consideration of parental preferences, and thus may be more likely to have delinquent friends. Delinquent peer groups, in turn, are a source of negative social capital, offering resources and rewards for misbehavior.
Hypothesis 3:
Along with the behavioral content of peer networks, the proportion of peers from networks of varying family structural (dis)advantages and the proportion of peers who are emotionally bonded to their parents are important for explaining violent delinquency. Invoking collective parenting arguments, we propose that peer groups with high proportions of adolescents from poor single-mother families increase the chances of violence in part because adolescents in the group are less likely to experience the benefits of bridging social capital (intergenerational interfamilial bonds). Adolescents embedded in networks with high proportions of friends from economically advantaged families that are not headed by single mothers, by contrast, are more likely to experience the benefits of intergenerational interfamilial bonds within the peer group. If youths from poor single-mother families associate with friends in similar circumstances, this increases the chances of being embedded in networks of shared family disadvantages, which can help explain the increased risk for violence among youths from poor single-mother families.
Hypothesis 4:
Hypothesis 4a: Embeddedness in networks with high proportions of friends from poor single-mother families increases violent delinquency and helps explain the effect on violence of living in a poor single-mother family.
Hypothesis 4b: Embeddedness in networks with high proportions of friends from nonpoor, non-single-mother families decreases violent delinquency and helps explain the effect on violence of living in a poor single-mother family.
The structural parenting potential in friendship networks likely reveals itself through shared emotional bonds between parents and adolescents in the network. In other words, the family structural characteristics of friendship networks may be relevant for violent offending because they influence the proportion of intergenerational bonds or bridging social capital in a network.
Hypothesis 5:
Data, Measurement, and Methods
Data
We assess our hypotheses using data from the Add Health, a nationally representative sample of seventh through twelfth graders randomly selected from 80 high schools and their feeder junior highs, 132 schools, in the U.S. during 1994 to 1995. We utilize data from the in-school and in-home interviews. The in-school interviews were administered to 90,118 students in 1994 and provide social network data based on responses to questions asking students to identify their best friends from a school roster with identification numbers that were used to link individuals together within networks. The in-school data also provides information on respondents’ home life, and behaviors.
Students who completed the in-school interview and those on the school enrollment rosters were eligible for the in-home surveys conducted nearly 1 year after the in-school survey. Approximately 200 students stratified by age and sex were randomly selected from each of the 80 pair schools, yielding a core sample of 12,105 adolescents. The in-home interviews were administered using computers to allow anonymous responses to questions on sensitive topics, including questions about violence. Our sample includes respondents who were attending school at the time of the initial in-home survey, were assigned valid sampling weights, and had no missing data on the study variables. This yields an analytic sample of 10,608 students.
Measures
Our substantive arguments are diagrammed in Figure 1. The first block of variables includes the family structural (dis)advantage and control variables. These variables are from the in-school surveys. The combined poverty and single-mother status of households captures family structural disadvantages and is measured with four categories: residence in a household that is neither poor nor headed by a single mother, residence in a poor household that is not headed by a single mother, residence in a nonpoor single-mother household, and residence in a poor single-mother household.
Substantive diagram.
Adolescents are coded as residing in a single-mother household if they reported that they reside with their biological mother, stepmother, foster mother, or adoptive mother and do not reside with their biological father, stepfather, foster father, or adoptive father. To qualify as a poor household, the residential parent must receive public assistance or both residential parents (or the single parent) must be in one of the following employment statuses with average salaries below the national living wage (U.S. Bureau of Labor Statistics, 2019): restaurant worker or personal service (waitress, housekeeper), transportation (bus driver, taxi driver), office support worker (bookkeeper, office clerk, secretary), does not work for pay, or has less than a high school diploma. A comparison of our poverty measure with a measure of parent reports of public assistance (social security, SSI, AFDC, food stamps, unemployment, or housing subsidies) shows considerable overlap, though our measure captures approximately 25% more impoverished families than the parent report of public assistance measure. This aligns with research showing that 22% of impoverished families do not receive public benefits (Edelstein et al., 2014). We also computed a more inclusive poverty measure from the parent interviews that codes a family as impoverished if they receive public assistance or report an income below $38,000 (the living wage adjusted to 1994 dollars). Sensitivity analyses in which we only included respondents in the analyses whose poverty responses from our measure matched each of the poverty measures we coded from the parent interview produce the same substantive results, yet the analytic sample size is reduced substantially.
In addition to our substantive variables, we control for race/ethnicity, age, sex, neighborhood disadvantage, urbanicity, and prior fighting. Race/ethnicity is coded as White, Black, Hispanic, Asian, and other from respondents’ self-reports of their race. For the small number of individuals who did not provide a racial classification (n = 18), race/ethnicity is assigned according to interviewer observation. Age is measured in years according to respondents’ self-reported date of birth. Sex is coded 0 for females and 1 for males. Community disadvantage is captured by a standardized summed scale of characteristics of the census tract in which adolescents reside, including the proportion of Black residents, female-headed households, divorced males, families below the poverty line, unemployed adults, and unemployed male adults (α = .90). Urban residence is coded 0 if the adolescent’s census block is partially rural and 1 if it is completely urbanized.
Our final control variable is a measure of fighting. Adolescents reported whether or not they got into a physical fight in the year preceding the in-school survey and are assigned a value of 1 if they reported getting into a physical fight and a value of 0 if they did not get into a fight. Inclusion of demographic, personal, community, and school variables helps account for selection effects, since these variables influence both peer networks and violence. Inclusion of prior violence in statistical models allows for a particularly stringent control for selection (Haynie & Osgood, 2005).
To capture bonding social capital, we include measures of parental supervision, monitoring, and emotional bonds. The in-school interviews do not offer comprehensive information on parenting, including only measures of emotional bonds. As such, we use parenting measures from the first wave in-home survey. For households with both parents, we use the parent who received the highest score. Although this does not capture parenting benefits of residing with two parents, it allows for comparison of single-mother versus other family forms and offers a conservative test of our arguments (see Demuth & Brown, 2004). We capture parental supervision using a mean index of three items assessing how often a parent is home when the adolescent leaves for school, returns from school, and goes to bed. Values for each item range from 0 (never) to 5 (always). Parental monitoring is a mean summed scale of adolescent reports of whether their parents make decisions for them about their curfew, bedtime, the people they hang around with, and television viewing. All responses are coded as 0 if parents do not make the decisions and 1 if they do. Parental bonds is a 3-item mean summed scale comprising adolescent reports on a 5-point scale (1 = strongly disagree to 5 = strongly agree) that they are satisfied with their communications with their parents, are satisfied with their relationship with their parents, and would describe their parents as warm and loving (α = .84). We conducted sensitivity analyses with a measure of parental bonds from the in-school survey asking adolescents how much their mother cares about them and how much their father cares about them (α = .90).
Our models also include four measures of peer networks. We combine information reported in the in-school interviews by respondents and their friends with social network nominations provided by respondents during the in-school survey to construct the peer network variables. Each respondent was asked to identify up to five male and five female closest friends from a roster of classmates. Using the rosters and questionnaire responses from nominated friends, we constructed network variables measuring the proportion of peers within each respondent’s friendship group who reside with a poor single mother, reside with a nonpoor non-single mother, and engaged in violent fights. We also captured the mean level of parent-child bonds within the network.
Two of the network variables capture the structural element of bridging social capital. These measures use the same criteria discussed above for deciphering whether adolescents reside in a household that is neither poor nor headed by a single mother or reside in a poor household with a single mother. We assess these measures in separate models to capture hypothesized structural disadvantages in networks with high proportions of adolescents from poor single-mother homes and structural advantages afforded by networks with high proportions of adolescents from households that are neither poor nor headed by single mothers. We capture the process element of bridging social capital using the in-school parental bonds measure. Finally, we constructed a measure of the percentage of peers who reported they were involved in a physical fight in the previous year to capture negative social capital embedded in peer networks.
Our dependent variable is a measure of violence in the past year. We focus on violent delinquency because prior studies show stronger correlations between poverty, single mothering, and offending when more serious forms of offending are considered. Our measure comprises four items asking respondents to provide anonymous reports during the Wave I in-home interview of whether in the past year they had been in a serious physical fight, pulled a knife or gun on someone, stabbed or shot someone, or hurt someone badly enough to need bandages or care from a doctor or nurse (α = .66). Because there is considerable overlap in the violence items, and multiple affirmative responses may in fact refer to the same incident, our dependent variable is a binary response of whether the respondent had engaged in any of the activities. The use of a binary measure aligns with our substantive interest in participation in serious fights or violence of any form. The measure we use in our presentation of results is similar to prior studies of violence using Add Health data (Haynie & Payne, 2006). We conducted sensitivity analyses with the dependent variable being only the serious fighting item to exactly match our control for prior violence and our measure of peer violence from the school interviews. The pattern of findings from these models are the same as those reported.
Analytic Strategy
We estimate a series of logistic regression models to predict violence captured during the in-home survey. Our models include covariates drawn primarily from the in-school survey, administered approximately 1 year prior to the in-home survey. We use parenting variables from the first wave in-home survey because the in-school survey did not provide information on parental supervision or monitoring. Sensitivity analyses using the in-school parental bond variable produce the same substantive results as those reported. Additional analyses using the dependent variable measured during the second in-home interview also affirm the robustness of the reported findings.
To assess the extent to which bonding, bridging, and negative social capital explain the effect of residing in an impoverished single-mother home, we compute indirect effects and standard errors using STATA’s GSEM command wherein we specify the outcome variable as dichotomous and mediating variables as continuous. All analyses take into consideration the clustering of the sample and use weights to compensate for differences in the selection probabilities, response rates across subgroups, and chance fluctuations of the sample from known population totals (Tourangeau & Shin, 1999).
Results
Table 1 provides unweighted descriptive statistics. Consistent with national averages, 21% of the adolescents in our study reside in single-mother households, and these adolescents are nearly twice as likely to experience poverty when compared to their peers. Specifically, 42% of single-mother families qualify as poor as compared with 20% of non-single mother families.
Unweighted Descriptive Statistics (N = 10,608).
Results from logistic regression models are presented in Table 2. Model 1 assesses the total effect of family structural disadvantage on violence with controls included in the model. Consistent with hypothesis 1, adolescents from households that are neither poor nor impoverished are less likely to engage in violence than those from the reference category, poor single-mother families (column 1, row 2). Similarly, adolescents from both poor and nonpoor families that are not headed by a single mother are less likely to engage in violence than those from poor single-mother families (column 1, rows 1 and 3). Z-test comparisons of the non-reference group family categories offer additional insight into the relationship between family structural (dis)advantages and violence. Comparisons of adolescents from single-mother homes that are not poor with those from poor non-single-mother homes (z = .300) and homes that are neither poor nor headed by single mothers (z = .364) show that those residing with single mothers with economic resources are not more likely to engage in violence than those from families that are not headed by single mothers. Thus, it is not single mothering per se that is associated with juvenile violence, but rather single mothering in the context of poverty.
Logistic Regression of Violence.
p < .05. **p < .01. ***p < .001 (two-tailed).
The next model includes bonding social capital variables and shows that, consistent with hypothesis 2, close relationships with parents are linked to reduced violence (column 2, row 16). Indirect effects presented in Table 3 show that the total effect of family structural disadvantage on fighting is mediated partially by parental bonds. That is, the indirect effects on violence of residing in a poor single-mother home relative to residing in other household forms are statistically significant (rows 1-3). Inconsistent with our second hypothesis, neither parental supervision nor parental monitoring significantly reduces violence (Table 2, column 2, rows 14 and 15). It may be that adolescents who are emotionally bonded to their parents have internalized parental messages, allowing successful parental control to occur through consideration of their parents’ reactions to misbehavior. In this case, direct supervision and monitoring become less important.
Generalized Structural Equation Indirect Effects on Violence.
p < .05. **p < .01. ***p < .001 (two-tailed).
Models 3 and 3a in Table 2 include variables capturing structural bridging social capital within peer networks (embeddedness in poor single-mother networks and in nonpoor non-single-mother networks). Embeddedness in networks with high proportions of friends from poor single-mother homes contributes to violence (column 3, row 17). This is an important finding as it supports our argument that peer networks are a proximal context in which collective parenting operates. The indirect effects in Table 3 show that the family disadvantage composition of peer networks helps explain the total effect of residing in a poor single-mother home relative to residing in other types of homes (rows 4–6). In other words, adolescents from poor single-mother homes are more likely to become embedded in networks of adolescents from similar family backgrounds, which proves relevant for understanding violence. This may be because of reduced collective parenting and less bridging social capital within structurally disadvantaged networks.
Model 3a in Table 2 presents a model that is similar to Model 3 but with consideration of family structural advantages built into peer networks. Our findings indicate that adolescents with higher proportions of friends who reside in households that are not headed by poor single mothers benefit from shared parental controls (column 5, row 17). The indirect effects in Table 3 show that family structural advantages within peer networks partially explain the total effect of residing in poor single-mother homes relative to residing in nonpoor families, whether they are female-headed or not (Table 3, rows 8 and 9). However, differences between adolescents from poor non-female-headed families and poor female-headed families are not associated with shared family structural advantages in peer networks (Table 3, row 7) because adolescents from these households are equally likely to be a part of advantaged networks.
Models 4 and 4a in Table 2 include measures of the proportion of adolescents’ friends who engage in fighting behaviors as well as the mean level of parent-child bonds within adolescents’ networks. In both models, adolescents who have a higher proportion of friends who fight also fight more themselves, as predicted by hypothesis 3 (columns 4 and 6, row 19). The indirect effects in Table 3, however, show that violent peer networks do not help explain the relationship between dwelling in an poor single-mother home and violence (rows 10–12), nor do they explain the controlling effects of parental bonds on violence (row 13).
Consistent with our proposition about process social capital in hypothesis 5, adolescents who interact with networks of peers who are emotionally bonded to their parents are less likely to engage in violence (Table 2, columns 4 and 6, row 20). The indirect effects in Table 3 show that being embedded in networks of peers who are close to their parents mediates part of the effect of living in poor single-mother homes as well as in homes that are not headed by single mothers (rows 14 and 16). Having peers who are bonded to their parents, however, does not contribute to differences in violence between adolescents from poor versus nonpoor single-mother homes (row 15). This is because adolescents from these types of homes are equally likely to select into friendship networks characterized by weak parent-child bonds.
Finally, Table 3 shows that networks of peers who are emotionally bonded with parents partially explain the effects of bonding social capital and structural bridging capital on violence. In particular, when adolescents are close to their parents, they are more likely to select friends who are similarly bonded with their parents, which in turn is associated with a reduction in violent delinquency (row 17). The family structural characteristics of networks appear to operate also partially through parental bonds in networks (rows 18 and 19).
Conclusions
Our research shows that it is important to differentiate single-mother households based on their poverty status in understanding violent delinquency. We find that adolescents from poor single-mother homes are more likely to engage in fighting behaviors than their peers who dwell in other types of households, including non-impoverished single-mother households and nonsingle-mother households. Adolescents who reside in single-mother homes that do note experience poverty are not at elevated risk for violence when compared to adolescents residing in non-single-mother homes. This is an important finding that speaks to debates about single motherhood and child outcomes and lends merit to recent calls to consider that it is not single-mothering per se that is relevant for understanding child and adolescent outcomes but rather the poverty that often accompanies single mothering. We also find support for our argument that bonding and bridging social capital help explain the elevated likelihood of violence among adolescents in impoverished single-mother families. Our findings suggest that if single mothers have economic and social resources to assist in carrying the burden of balancing finances and family, they are able to foster emotional relationships with their children that help control violence among both their own children and their children’s close friends.
This latter point is highlighted by our findings pertaining to the structural and process parenting potential residing within friendship networks. Criminologists who study communities and crime have long suggested that collective parenting within disadvantaged neighborhoods is undermined by the proportion of single mothers residing in the area (Sampson, 2012). We advance this argument substantively and empirically by focusing on the relatively intimate setting of peer groups as a resource for bridging capital. In doing so, we find that family structural (dis)advantages within peer groups, as well as intergenerational bonds between parents and youth in peer groups, are important for understanding juvenile violence generally and for understanding why youths living with impoverished single-mothers are at elevated risk for violent delinquency.
Although negative social capital (embeddedness in networks with a high proportion of violent peers) is a strong predictor of violence, it is not relevant for linking family structural disadvantages to violence. This is an interesting finding because it challenges the singular focus on peers as sources of negative social capital in prior studies of family disadvantages and juvenile offending. Our focus on bridging social capital in friendship networks emphasizes that peer groups can serve not only as sources of negative social capital that promote violence, but also as sources of bridging social capital that control violence. Our research thereby offers promising new lines of inquiry for research on how peers matter in studies of family structure, family poverty, and adolescent violence. Specifically, we maintain that continued research on parenting within youths’ families as well as within their friends’ families is important for pushing forward studies of adolescent behaviors.
Given that our research is not without limitations, we envision many ways in which future research can push beyond our initial findings regarding bridging social capital within peer networks. For instance, future research on bridging social capital in peer networks should consider the myriad family structures and living arrangements in which adolescents come of age in contemporary society, including examinations of shared physical custody of children, step parenting, and additional family forms experienced by youths and their networks of friends. Continued research also could examine within-individual changes in family forms over the life-course and the timing of family structure changes. For instance, children who were born to single mothers may select into friendship groups differently than those who faced more recent household changes (Cherlin, 2009). As families continue to shift and adapt to economic and cultural shifts, both positive and negative, it becomes increasingly relevant to consider how parenting within households and collectively within adolescent friendship groups can ease family transitions, family adaptations, and alternative ways of doing family in ways that encourage adolescents to refrain from offending behaviors. New surveys of self-reported delinquency may consider including questions to allow such analyses.
A particularly important avenue for future research would be to offer more thorough consideration of precisely how and why bridging social capital is related to adolescent violence. Although our collective parenting argument emphasizes the relevance of parental controls (in the form of emotional bonds, supervision, and monitoring), it may be the case that parents from advantaged households also offer their children’s less advantaged peers access to additional conventional resources, networks, and cultural capital that enable them to meet identity, status, and respect needs through legitimate means (Lareau, 2011), rather than through violence (Anderson, 1999; Jones, 2009; Rios, 2011). Researchers have begun to embed qualitative components into large-scale quantitative studies (see Newman & Massengill, 2006). A focus on collective parenting in these data collection efforts could open avenues for considering precisely how the advantages or disadvantages of families in adolescents’ peer networks accrue in ways relevant for understanding delinquency.
Finally, we encourage research on the influence of bridging social capital within peer networks on additional outcomes. Our study focused on adolescent violence with a relatively limited measure of participation in violence. There is a good possibility that the family experiences of youths’ friends and the resulting bridging social capital may influence the frequency with which adolescents use violence, as well as additional forms of offending, including property and drug offenses. Consideration of positive outcomes, such as school performance and educational attainment, would provide additional insights into how bridging social capital operates within friendship networks.
Overall, our study significantly extends research on single-mothering and violent delinquency. First, we show that living in a single-mother home is relevant for adolescent violence only in the context of poverty. Second, we broaden understanding of social capital resources within and between families that set the stage for the observed link between single-mothering in poverty and adolescent violence. In doing so, we highlight the role of the family experiences and resources available to youths’ friends, as well as their own family resources and experiences. Our findings are consistent with the spirit of studies of communities and crime and collective parenting, yet they extend existing studies to highlight the relevance of friendship groups as a domain for collective parenting. In short, understanding connections between family disadvantages and youth violence requires considering the family experiences of youths’ friends, above and beyond the family experiences of youths themselves.
Footnotes
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) received no financial support for the research, authorship, and/or publication of this article.
